This paper surveys the neoclassical theory of aggregate investment and its criticisms. We distinguish four main formulations of this theory: the traditional ‘Wicksellian’ investment function; the Fisherian ‘array-of-opportunities’ approach; the Jorgensonian model; the now prevailing adjustment-costs models.

With respect to other papers criticizing the neoclassical theory of investment, we do not appeal to market imperfections. We instead argue that all four formulations present serious theoretical difficulties, even conceding free competition. The derivation of a negative relation between investment and the interest rate relies on the traditional conception of capital as a single homogeneous factor, that has been proved untenable by the so-called Cambridge capital controversies. Marginalist factor substitution mechanisms, on which the theory is based, have been shown to be deprived of theoretical foundations. Old and new attempts to derive the interest-elastic investment function without making recourse to factor substitution (as in the so-called ‘array-of-opportunities’ approach or in models with convex adjustment costs) are even less solid, because they rest on the mistaken assumption that expected rates of return on investment projects are independent from the interest rate.

In addition to this central problem, that affects both traditional and contemporary neoclassical theory, we will discuss some additional issues raised by contemporary models. The dominant approach, nowadays, is to derive demand for capital from the Jorgenson (1963) model, focusing on an atomistic price-taking representative firm, and append an adjustment process based on some assumption regarding the internal costs of adjustment. We will discuss the difficulties that this approach encounters in determining the optimal capital stock under constant returns to scale, and the aggregation problems that arise when the investment function of the single firm is applied to the whole economy.